WhatsApp AI or ChatGPT: which is better?

All about ChatGPT

All about Artificial intelligence

All about WhatsApp

WhatsApp's AI arrived on the messaging app in October this year, which brings comparisons between Meta AI and ChatGPT. The latter uses the GPT family language model, Meta AI uses Llama 3. Both are modern, but differences persist between the two AIs

WhatsApp AI or ChatGPT: who provides the best answers?

ChatGPT provides better responses and makes fewer errors in responses than Meta IA. But both tools have hallucinations and can invent facts. Users reported that Meta IA already considered that 9.11 is greater than 9.8, which is not true, because the latter corresponds to 9.80.

On the other hand, a study indicated that ChatGPT gets more than half of programming questions wrong. According to this research, 52% of ChatGPT responses were inaccurate and another 77% were unnecessarily long. Another study found that the tool also gives incorrect and inaccurate answers about medicines.

In addition to text commands, ChatGPT also understands images and audio, but WhatsApp's AI is trained only for texts.

This makes OpenAI's AI offer more options, as it can, for example, help identify elements in photos and answer questions about maps and graphs. The user can also upload tables and PDFs and request a summary of them from ChatGPT.

ChatGPT is also better than WhatsApp's AI at solving difficult problems, as it can process complex tasks that require a reasonable level of logic.

Images

On the other hand, it is possible to obtain AI-generated images for free through Meta IA. In the case of ChatGPT, the AI ​​that generates the images is Dall-e, whose free version has a limit on the number of images that can be produced.

Also read:

Data security and privacy

Whether in WhatsApp AI or ChatGPT, security arouses distrust among users and experts.

In March 2023, an issue allowed users to see each other's conversation history. The bug also caused the leakage of payment information for around 1.2% of ChatGPT subscribers, according to a report from OpenAI.

ChatGPT saves conversations, which are reviewed and used by OpenAI for model training purposes, as the company itself indicates in the FAQ. But the user can choose not to have their chat used by the company on the privacy portal.

The company informs that it cannot delete commands from the history and, therefore, the user should not share sensitive information. According to consultancy Kaspersky, the potential for data misuse is a valid concern.

Like ChatGPT, Meta clarified that WhatsApp AI user data will be used to train the tool. Despite the content of the conversation being used, the company states that no information will be recorded by the system.

According to Meta, the tool does not have access to messages and personal calls. But the company's privacy policy allows the use of metadata, such as location, cell phone type and version of the user's application.